sinzlab/cGNF

This is the official code for calibration in multi-hypothesis human pose estimation

21
/ 100
Experimental

This project helps researchers and developers working on analyzing human movement to refine their 3D pose estimation models. It takes raw video data or 2D keypoints of human motion as input and outputs a more accurate, calibrated 3D representation of human poses. It's designed for someone specializing in computer vision, biomechanics, or sports science who needs precise, multi-hypothesis 3D human pose data.

No commits in the last 6 months.

Use this if you are developing or evaluating 3D human pose estimation models and need to ensure their probabilistic outputs are well-calibrated and accurately reflect real-world pose uncertainty.

Not ideal if you are looking for a simple, out-of-the-box solution for general 3D pose tracking in real-time applications without deep dives into model calibration.

human-pose-estimation 3d-motion-analysis computer-vision-research biomechanics probabilistic-modeling
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

9

Forks

Language

Jupyter Notebook

License

MIT

Last pushed

May 30, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/sinzlab/cGNF"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.